Summary of Box2flow: Instance-based Action Flow Graphs From Videos, by Jiatong Li et al.
Box2Flow: Instance-based Action Flow Graphs from Videos
by Jiatong Li, Kalliopi Basioti, Vladimir Pavlovic
First submitted to arxiv on: 30 Aug 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper presents an instance-based method called Box2Flow for predicting accurate and rich flow graphs from a single procedural video. The proposed approach extracts bounding boxes, predicts pairwise edge probabilities between step pairs, and builds the flow graph using a spanning tree algorithm. This method can be used to capture detailed step descriptions in task-based methods. Experimental results on MM-ReS and YouCookII datasets show that Box2Flow effectively extracts flow graphs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper helps computers learn how to understand videos that show people doing things, like cooking or fixing things. These videos usually have a specific order of steps, but the same step can be done in different ways. The goal is to make a picture that shows how these steps relate to each other. Current methods try to create one big picture for all videos of the same task, but they don’t capture the details very well. This paper proposes a new way to learn these pictures from just one video. It works by finding boxes around important parts of the video and then connecting them in a special order. The results show that this method can make good pictures. |
